I went through all the questions here and discussion available on web, to figure out how the 'scale_pos_weight' (and 'class_Weight' for multi class problems & individual instance weights given by the 'weight' param) impact the calculations of the loss gradient / other calculations, but without much luck. Can someone please explain how this params impact the loss calculations, or any other calculations in xgboost/lightgbm.

Also, I read (somewhere) that weighting using the scale_pos_weight, heavily impacts the intercept in the model (read similar is case for logistic regression, where weighting the minor class more impacts the intercept heavily). Can someone explain how.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.